基于点云片段匹配约束和轨迹漂移优化的回环检测方法
Segment Based Loop Detection with point cloud matching constraint And Trajectory Drift Optimization

基于三维点云的同时定位与建图(Simultaneous Localization and Mapping, SLAM)是机器人导航与定位领域重要的技术之一. 然而具有回环检测功能的这类SLAM系统仍鲜见于文献中. 本文首先提出了一种新的基于三维点云的室外SLAM 系统的框架, 该框架由里程计, 回环检测, 位姿优化三部分组成. 其次针对回环检测，我们提出一种基于点云片段匹配约束的方法提升回环检测的效率. 最后针对位姿优化, 我们提出两种轨迹漂移优化算法 分别为全局一致性的回环调整算法和位姿预测和补偿算法. 我们通过广泛的实验验证本文提出的方法, 结果表明本文所提出的SLAM 系统具有稳定和精确的位姿估计能力.
Simultaneous Localization and Mapping (SLAM) based on point cloud is one of the important technologies in robot navigation and positioning. However, this type of SLAM system with loop detection is still rare in the literature. In this paper we first propose a new point cloud-based outdoor SLAM system framework consisting of three parts: odometry, loop detection, and pose optimization. Second, we develop a method based on point cloud segment matching constraints to improve the efficiency of loop detection. Finally, we present two trajectory drift optimization methods, including Global consistency loop adjustment algorithm and pose prediction and compensation algorithm. We have validated the proposed method through extensive experiments. The results show that the proposed SLAM system has stable and accurate pose estimation capabilities.